# US 12,223,725 B2 · Object identification method and related surveillance system > Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt - Number: US 12,223,725 B2 - Document ID: US12223725B2 - Status: Granted - Kind: B2 - Application number: 17/537,390 - Filed: 2021-11-29 - Published: 2022-06-16 - Granted: 2025-02-11 - Earliest claimed: 2020-12-10 - Assignees: Vivotek Inc. - Applicants: Vivotek Inc. - Inventors: Cheng-Chieh Liu - Web page: https://patentgrep.com/patent/US12223725B2 ## CPC - G06V 20/52: Surveillance or monitoring of activities, e.g. for recognising suspicious objects (recognising microscopic objects G06V20/69) - G06V 10/761: {Proximity, similarity or dissimilarity measures} - G06V 10/762: using clustering, e.g. of similar faces in social networks - G06V 10/772: Determining representative reference patterns, e.g. averaging or distorting patterns; Generating dictionaries - G06V 10/774: Generating sets of training patterns; Bootstrap methods, e.g. bagging or boosting - G06V 40/20: Movements or behaviour, e.g. gesture recognition (recognition of facial expressions G06V40/16) - G06V 40/23: {Recognition of whole body movements, e.g. for sport training} - G06V: IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING ## Abstract An object identification method is applied to a surveillance system. The surveillance system includes at least one surveillance apparatus. The object identification method includes acquiring a plurality of first feature vectors of a first moving object and a plurality of second feature vectors of at least one second moving object within a series of surveillance images from the surveillance apparatus, transforming the first feature vectors and the second feature vectors respectively into a first cluster distribution set and at least one second cluster distribution set, comparing similarity of the first cluster distribution set and the at least one second cluster distribution set, and setting a similarity ranking of the first cluster distribution set and the at least one second cluster distribution set according to a comparison result so as to determine whether the first moving object and the at least one second moving object are the same. ## Claims ### Claim 1 An object identification method applied to a surveillance system, the surveillance system having at least one surveillance apparatus, the object identification method comprising: acquiring a plurality of first feature vectors of a first moving object and a plurality of second feature vectors of at least one second moving object within a series of surveillance images from the surveillance apparatus; transforming the plurality of first feature vectors into a first cluster distribution set, so as to record a first appearing time interval and a first cluster center of the first cluster distribution set, and a first specific point of time representing the first cluster distribution set; transforming the plurality of second feature vectors into at least one second cluster distribution set; applying a main similarity comparison to the first cluster distribution set and the at least one second cluster distribution set; and setting a similarity ranking of the at least one second cluster distribution set relative to the first cluster distribution set in accordance with a main comparison result, so as to determine whether the first moving object and the at least one second moving object are the same. ### Claim 2 (depends on 1) The object identification method of claim 1, wherein acquiring the plurality of first feature vectors of the first moving object within the series of surveillance images comprises: acquiring a first trace datum of the first moving object; searching a plurality of surveillance images relevant to the first trace datum from the series of surveillance images; and analyzing the plurality of surveillance images to acquire the plurality of first feature vectors. ### Claim 3 (depends on 1) The object identification method of claim 1, wherein acquiring the plurality of first feature vectors of the first moving object within the series of surveillance images comprises: dividing each of the series of surveillance images into an alpha area and a beta area in accordance with the first moving object; computing a first alpha area sub-feature vector of the first moving object in the alpha area and a first beta area sub-feature vector of the first moving object in the beta area within the surveillance image relevant to a trace datum of the first moving object; and combining the first alpha area sub-feature vector and the first beta area sub-feature vector to set as a corresponding first feature vector of the surveillance image. ### Claim 4 (depends on 3) The object identification method of claim 3, wherein each surveillance image is uniformly divided into the alpha area and the beta area in accordance with the first moving object, or a region of interest is set in each surveillance image to be one of the alpha area and the beta area. ### Claim 5 (depends on 3) The object identification method of claim 3, further comprising: transforming a plurality of first alpha area sub-feature vectors of the first moving object into a first alpha area cluster distribution set; transforming a plurality of second alpha area sub-feature vectors of the at least one second moving object into at least one second alpha area cluster distribution set; applying a sub similarity comparison for the at least one second alpha area cluster distribution set and the first alpha area cluster distribution set; and resetting the similarity ranking acquired by the main similarity comparison via a sub comparison result. ### Claim 6 (depends on 1) The object identification method of claim 1, wherein some corresponding first feature vectors relevant to each point of time within the first appearing time interval are acquired, and one of the foresaid corresponding first feature vectors closest to the first cluster center is defined as a specific first feature vector, and a point of time for the specific first feature vector is set as the first specific point of time. ### Claim 7 (depends on 1) The object identification method of claim 1, further comprising: acquiring a third cluster distribution set of a third moving object and a fourth cluster distribution set of a fourth moving object within the series of surveillance images; determining appearing time intervals of the third cluster distribution set and the fourth cluster distribution set are overlapped with the first appearing time interval of the first cluster distribution set; applying the main similarity comparison for the first cluster center and cluster centers of the third cluster distribution set and the fourth cluster distribution set; and displaying two surveillance image respectively relevant to the third cluster distribution set and the fourth cluster distribution set and acquired at specific points of time alongside a correspond surveillance image relevant to the first cluster distribution set and acquired at the first specific point of time in accordance with the main comparison result. ### Claim 8 (depends on 1) The object identification method of claim 1, further comprising: acquiring a fifth cluster distribution set of a fifth moving object adjacent to the first moving object and a sixth cluster distribution set of a sixth moving object adjacent to the second moving object within the series of surveillance images; applying another main similarity comparison for the fifth cluster distribution set and the sixth cluster distribution set; and setting another similarity ranking of an assembly of the first moving object and the fifth moving object and an assembly of the second moving object and the sixth moving object in accordance with two main comparison results. ### Claim 9 A surveillance system, comprising: at least one surveillance apparatus adapted to acquire a series of surveillance images; and an operation processor electrically connected to the at least one surveillance apparatus in a wire manner or in a wireless manner, the operation processor being adapted to acquire a plurality of first feature vectors of a first moving object and a plurality of second feature vectors of at least one second moving object within the series of surveillance images, to transform the plurality of first feature vectors into a first cluster distribution set for recording a first appearing time interval and a first cluster center of the first cluster distribution set, and a first specific point of time representing the first cluster distribution set, to transform the plurality of second feature vectors into at least one second cluster distribution set, to apply a main similarity comparison to the first cluster distribution set and the at least one second cluster distribution set, and set a similarity ranking of the at least one second cluster distribution set relative to the first cluster distribution set in accordance with a main comparison result, so as to determine whether the first moving object and the at least one second moving object are the same. ### Claim 10 (depends on 9) The surveillance system of claim 9, wherein the operation processor is further adapted to acquire a first trace datum of the first moving object, to search a plurality of surveillance images relevant to the first trace datum from the series of surveillance images, and to analyze the plurality of surveillance images to acquire the plurality of first feature vectors. ### Claim 11 (depends on 9) The surveillance system of claim 9, wherein the operation processor is further adapted to divide each of the series of surveillance images into an alpha area and a beta area in accordance with the first moving object, to compute a first alpha area sub-feature vector of the first moving object in the alpha area and a first beta area sub-feature vector of the first moving object in the beta area within the surveillance image relevant to a trace datum of the first moving object, and to combine the first alpha area sub-feature vector and the first beta area sub-feature vector to set as a corresponding first feature vector of the surveillance image. ### Claim 12 (depends on 11) The surveillance system of claim 11, wherein each surveillance image is uniformly divided into the alpha area and the beta area in accordance with the first moving object, or a region of interest is set in each surveillance image to be one of the alpha area and the beta area. ### Claim 13 (depends on 11) The surveillance system of claim 11, wherein the operation processor is further adapted to transform a plurality of first alpha area sub-feature vectors of the first moving object into a first alpha area cluster distribution set, to transform a plurality of second alpha area sub-feature vectors of the at least one second moving object into at least one second alpha area cluster distribution set, to apply a sub similarity comparison for the at least one second alpha area cluster distribution set and the first alpha area cluster distribution set, and to reset the similarity ranking acquired by the main similarity comparison via a sub comparison result. ### Claim 14 (depends on 9) The surveillance system of claim 9, wherein some corresponding first feature vectors relevant to each point of time within the first appearing time interval are acquired, and one of the foresaid corresponding first feature vectors closest to the first cluster center is defined as a specific first feature vector, and a point of time for the specific first feature vector is set as the first specific point of time. ### Claim 15 (depends on 9) The surveillance system of claim 9, wherein the operation processor is further adapted to acquire a third cluster distribution set of a third moving object and a fourth cluster distribution set of a fourth moving object within the series of surveillance images, to determine appearing time intervals of the third cluster distribution set and the fourth cluster distribution set are overlapped with the first appearing time interval of the first cluster distribution set, to apply the main similarity comparison for the first cluster center and cluster centers of the third cluster distribution set and the fourth cluster distribution set, and display two surveillance image respectively relevant to the third cluster distribution set and the fourth cluster distribution set and acquired at specific points of time alongside a correspond surveillance image relevant to the first cluster distribution set and acquired at the first specific point of time in accordance with the main comparison result. ### Claim 16 (depends on 9) The object identification method of claim 9, wherein the operation processor is further adapted to acquire a fifth cluster distribution set of a fifth moving object adjacent to the first moving object and a sixth cluster distribution set of a sixth moving object adjacent to the second moving object within the series of surveillance images, to apply another main similarity comparison for the fifth cluster distribution set and the sixth cluster distribution set, and to set another similarity ranking of an assembly of the first moving object and the fifth moving object and an assembly of the second moving object and the sixth moving object in accordance with two main comparison results.